Data (2)

Ways of ensuring good quality data from the proposal and claims forms

Questions should be well-designed and unambiguous so that full information is given and so that applications/ claims can be easily processed.

Use questions with quantitative or tick-box answers wherever possible

Questions should be in the same order as the input into the administration systems, for quick processing of applications/ claims.

Ask the policyholder to verify a copy of the key information

Need for the proposal form at the time of the claim

All rating factors required should be readily identifiable (on the proposal form) so that the composition of the final premium can be determined

Underwriting results should be added to the proposal form

Forms should be designed so that information can be easily analysed and cross-checks made between the two sources

To help check the validity of the claim

to update policy information, e.g. the policyholder has died/ total loss under general insurance

Importance of retaining past policy and claim records

When an insurance company analyses past experience in order to help set future assumptions, several past years' worth of data are often needed in order to give a sufficient volume of data, or to identify trends.

Problem with data for employee benefit schemes

The actuary does not have full control over the data, as it is provided by the sponsor

May result in poor quality or summarised data

Therefore particularly important to validate this type of data

Sources of data for a valuation of a benefit scheme

Membership data

Data from previous valuation

Accounting data

Full listing of the actual assets held

Reconciliation checks

Reconciling the total number of members/ policies and changes in membership/ policies using previous data and movement data

Reconciling the total benefit amounts and premiums and changes in them, using previous data and movement data

Where assets are held by a third party, reconciliation between the beneficial owner's and the custodian's records

Reconciling shareholdings at the start and end of the period, adjusted for sales and purchases, and bonus issues

Cross-checks

Checking movement data against accounting data, e.g. benefit payments

Checking membership data against accounting data, e.g. contributions

Checking asset data against accounting data, e.g. investment returns

Full deed audit, for example checking title deeds to large real property assets

Reasonableness checks

Checking the average sum assured or premium looks sensible for class of business

Checking the average sum assured or premium against previous data

Checking for unusual values, impossible dates or missing records

Spot checks

Random checking of individual member or policy records

Checking individual assets or liabilities exist/ are held on a given date

Checking that the correct value of an asset or liability has been recorded

Problems with summarised data

The reliability of the valuation will be reduced, as full validation of the data is impossible.

Summarised data may miss significant differences between the nature of the benefits that have been grouped together, e.g. the structure of the membership may have changed

Summarised data cannot be used to value options and guarantees that apply at an individual level.

Summarised data is therefore only suitable if such inaccuracies are recognised by the users of the results based on the data.

Examples of industry-wide collection schemes in the UK

Association of British Insurers (wide variety of insurance data)

Continuous Mortality Investigation Bureau of the IFoA (mortality and morbidity data)

Reasons why industry data is not directly comparable

Different geographical or socio-economic markets

Different policies (i.e. cover, terms and conditions)

Different sales methods

Different practices, e.g. underwriting and claims settlement processes

Different nature of data stored

Different coding of risk factors, e.g. definition of a smoker